WIP: Better suggest

common/ostream.hpp

@@ -54,6 +54,13 @@ namespace acommon {
       return *this;
     }
 
+    OStream & operator << (int i) {printf("%d", i); return *this; }
+    OStream & operator << (unsigned i) {printf("%u", i); return *this;}
+    OStream & operator << (long int i) {printf("%ld", i); return *this; }
+    OStream & operator << (long unsigned i) {printf("%lu", i); return *this;}
+    OStream & operator << (long long int i) {printf("%lld", i); return *this; }
+    OStream & operator << (long long unsigned i) {printf("%llu", i); return *this;}
+
     OStream & operator << (ParmStr in) {
       write(in);
       return *this;

common/string_pair.hpp

@@ -7,10 +7,8 @@
 #ifndef ASPELL_STRING_PAIR__HPP
 #define ASPELL_STRING_PAIR__HPP
 
-
 namespace acommon {
 
-
 struct StringPair {
   const char * first;
   const char * second;
@@ -19,7 +17,10 @@ struct StringPair {
   StringPair() : first(""), second("") {}
 };
 
-
 }
 
+#ifdef __clang__
+#pragma clang diagnostic ignored "-Wreturn-type-c-linkage"
+#endif
+
 #endif /* ASPELL_STRING_PAIR__HPP */

modules/filter/markdown.cpp

@@ -554,31 +554,31 @@ struct HtmlTag : MultilineInline {
     switch (state) {
       // note: this switch is being used as a computed goto to make
       //   restoring state straightforward without restructuring the code
-    case Between:
+     case Between:
       if (asc_isalpha(*itr) || *itr == '_' || *itr == ':') {
         itr.inc();
         while (asc_isalpha(*itr) || asc_isdigit(*itr)
                || *itr == '_' || *itr == ':' || *itr == '.' || *itr == '-')
           itr.inc();
-      case AfterName:
+       case AfterName:
         itr.eat_space();
         if (itr.eol()) return AfterName;
         if (*itr != '=') return Invalid;
         itr.inc();
-      case AfterEq:
+       case AfterEq:
         itr.eat_space();
         if (itr.eol()) return AfterEq;
         if (*itr == '\'') {
           itr.inc();
-        case InSingleQ:
+         case InSingleQ:
           while (!itr.eol() && *itr != '\'')
             itr.inc();
           if (itr.eol()) return InSingleQ;
           if (*itr != '\'') return Invalid;
           itr.inc();
         } else if (*itr == '"') {
           itr.inc();
-        case InDoubleQ:
+         case InDoubleQ:
           while (!itr.eol() && *itr != '"')
             itr.inc();
           if (itr.eol()) return InDoubleQ;
@@ -594,12 +594,14 @@ struct HtmlTag : MultilineInline {
         }
         return Between;
       }
-    case BeforeClose:
+     case BeforeClose:
       return BeforeClose;
-    case Valid: case Invalid:
+     default: //case Valid: case Invalid:
       // should not happen
-      abort();
+      break;
     }
+    // should not be here
+    abort();
   }
 };
 

modules/speller/default/editdist.cpp

@@ -18,7 +18,7 @@ namespace aspeller {
     int a_size = a0.size() + 1;
     int b_size = b0.size() + 1;
     VARARRAY(short, e_d, a_size * b_size);
-    ShortMatrix e(a_size,b_size,e_d);
+    Matrix<short> e(a_size,b_size,e_d);
     e(0, 0) = 0;
     for (int j = 1; j != b_size; ++j)
       e(0, j) = e(0, j-1) + w.del1;

modules/speller/default/matrix.hpp

@@ -5,16 +5,17 @@
 
 namespace aspeller {
 
-  class ShortMatrix {
+  template <typename T>
+  class Matrix {
     int x_size;
     int y_size;
-    short * data;
+    T * data;
   public:
-    void init(int sx, int sy, short * d) {x_size = sx; y_size = sy; data = d;}
-    ShortMatrix() {}
-    ShortMatrix(int sx, int sy, short * d) {init(sx,sy,d);}
-    short operator() (int x, int y) const {return data[x + y*x_size];}
-    short & operator() (int x, int y) {return data[x + y*x_size];}
+    void init(int sx, int sy, T * d) {x_size = sx; y_size = sy; data = d;}
+    Matrix() {}
+    Matrix(int sx, int sy, T * d) {init(sx,sy,d);}
+    T operator() (int x, int y) const {return data[x + y*x_size];}
+    T & operator() (int x, int y) {return data[x + y*x_size];}
   };
 
 }

modules/speller/default/suggest.cpp

@@ -198,6 +198,8 @@ namespace {
     ObjStack           buffer;
     ObjStack           temp_buffer;
 
+    String             str_buffer;
+
     static const bool do_count = true;
     static const bool dont_count = false;
 
@@ -303,6 +305,7 @@ namespace {
       return (word_weight*word_score
               + soundslike_weight*soundslike_score)/100;
     }
+    int nudge_score(ParmStr orig_norm, int score, ParmStr normalized, ParmStr word);
     int skip_first_couple(NearMisses::iterator & i) {
       int k = 0;
       InsensitiveCompare cmp(lang);
@@ -1263,12 +1266,12 @@ namespace {
       adj_threshold = 0;
       unsigned int j;
 
-      CharVector orig_norm, word;
+      Vector<NormalizedChar> orig_norm, word;
       orig_norm.resize(original.word.size() + 1);
       for (j = 0; j != original.word.size(); ++j)
         orig_norm[j] = parms->ti->to_normalized(original.word[j]);
       orig_norm[j] = 0;
-      ParmString orig(orig_norm.data(), j);
+      NormalizedString orig_word_norm(orig_norm.data(), j);
       word.resize(max_word_length + 1);
 
       for (i = scored_near_misses.begin();
@@ -1283,10 +1286,12 @@ namespace {
           for (j = 0; (i->word)[j] != 0; ++j)
             word[j] = parms->ti->to_normalized((i->word)[j]);
           word[j] = 0;
-          int new_score = typo_edit_distance(ParmString(word.data(), j), orig, *parms->ti);
+          NormalizedString word_norm(word.data(), j);
+          int new_score = typo_edit_distance(word_norm, orig_word_norm, *parms->ti);
           // if a repl. table was used we don't want to increase the score
           if (!i->repl_table || new_score < i->word_score)
             i->word_score = new_score;
+          i->word_score = nudge_score(orig_word_norm.str(), i->word_score, word_norm.str(), i->word);
           i->adj_score = adj_wighted_average(i->soundslike_score, i->word_score, parms->ti->max);
         }
         if (i->adj_score > adj_threshold)
@@ -1308,6 +1313,17 @@ namespace {
     }
   }
 
+  int Working::nudge_score(ParmStr orig_norm, int score, ParmStr normalized, ParmStr word) {
+    int base_score = edit_distance(orig_norm, normalized, parms->edit_distance_weights);
+    const char * fixed_word = lang->fix_case(original.case_pattern, word, str_buffer);
+    int new_score = edit_distance(original.word, fixed_word, parms->edit_distance_weights);
+    //CERR.printf("%s %s %d >? %d\n", original.word.c_str(), fixed_word, new_score, base_score);
+    if (new_score > base_score)
+      return score + parms->ti->case_mismatch;
+    else
+      return score;
+  }
+
   void Suggestions::transfer(NearMissesFinal & near_misses_final) {
 #  ifdef DEBUG_SUGGEST
     COUT << "\n" << "\n"